Download Prediction of Breast Cancer Risk Level with Risk Factors in

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Breast milk wikipedia , lookup

Preventive healthcare wikipedia , lookup

Computer-aided diagnosis wikipedia , lookup

Risk factors for breast cancer wikipedia , lookup

Transcript
International Journal of Computer Applications (0975 – 8887)
Volume 82 – No4, November 2013
Prediction of Breast Cancer Risk Level with Risk Factors
in Perspective to Bangladeshi Women using Data Mining
Kawsar Ahmed
Md. Ahsan Habib
Department of Information and Communication
Technology, Mawlana Bhashani Science and
Technology University, Bangladesh
Department of Information and Communication
Technology, Mawlana Bhashani Science and
Technology University, Bangladesh
Tasnuba Jesmin
Md. Zamilur Rahman
Department of Information and Communication
Technology, Mawlana Bhashani Science and
Technology University, Bangladesh
Department of Information and Communication
Technology, Mawlana Bhashani Science and
Technology University, Bangladesh
Md. Badrul Alam Miah
Department of Information and Communication
Technology, Mawlana Bhashani Science and
Technology University, Bangladesh
ABSTRACT
The upgraded and modern medical technologies are the most
challenging task to detect cancer and provide accurate
treatment. In Bangladesh about two million women are
affected by 2nd most occurring deathful breast cancer due to
them and their family member’s unconsciousness and poverty.
It requires about $400-500 for proper diagnosis and treatment.
Most of the Bangladeshi women are uneducated and feel shy
with society or husband to go doctor for checking breast
cancer. So it also will be a good achievement of this work to
find breast cancer with more efficiency. Breast cancer
depends on some risk factors that may help to detect breast
cancer using multi-layered approach. In this work, at first it is
collected 100 peoples' information which consist of both
cancer and non-cancer information having missing or
duplicate information. So pre-processing and K-means
clustering methods are performed to separate relevant and
non-relevant data to Breast Cancer. Then risk factors are
ranked using WEKA tools and are assigned a score according
to rank. Finally, it is implemented an application software
using Lotus Notes to predict Breast Cancer risk level which is
easier, effective, efficient, secured, cheap and time saving
with some suggestions. This technique will contribute equal
opportunity to the underdeveloped and developing countries
to detect, diagnosis, and treatment of breast cancer.
General Terms
Computer Science, Data Mining, Breast Cancer in Bangladesh
Keywords
Breast cancer in Bangladesh, Public health, Data mining, Risk
factors of breast cancer, WEKA toolkit, Woman Health
Conditions
1. INTRODUCTION
Cancer is a class of most deathful diseases caused by out-ofcontrol cell growth. There are more than 100 different types
of cancer, and each is classified by the type of cell that is
initially affected. Cancer harms the body when damaged cells
by dividing uncontrollably. Cancer is treated as most harmful
disease in the world because most of the patients died when
affected properly. All of cancers breast cancer is the most in
Bangladesh and 2nd in the world deathful disease. Breast
cancer is a cancer that begins in the tissues of the breast.
There are two types of Breast cancer one is Ductal carcinoma
and another is Lobular carcinoma. Ductal carcinoma begins in
the tubes that move milk from the breast to the nipple. Most
breast cancers are of this type. It may progress to invasive
cancer if untreated. Lobular carcinoma starts in the parts of
the breast, called lobules, which produce milk.
World Health Organization (WHO) says that the death rate of
breast cancer is high in Bangladesh and ranked Bangladesh
2nd deathful country in the world for breast cancer of women.
More than 89% of Bangladeshis people are Muslim, ranked
3rd in the world after Indonesia and Pakistan. Approximately
45% of the population is related in the agricultural sector.
Bangladesh is a low-income country, defined by the World
Bank as countries with a Gross National Income (GNI) less
than US $1,005 per capita. About 40% of the population is
underemployed; many participants in the labour force work
only a few hours a week, at low wages. Approximately 60%
of women are illiterate, and 27% of the population is
undernourished. In Bangladesh only one doctor give service
more than 3,300 people for urban areas and more than 15,000
people for rural areas [1]. So it is a burning question of
Bangladesh about general population health.
Most of the women don’t know about breast cancer before
affected for a long period of time. Many of them suffered
from husband as well as society. They believe that breast
cancer is caused by an evil spirit, or as a punishment for bad
deeds. Even some families give women very low priority. So
it’s need to increase the awareness among all population both
rural and urban areas in Bangladesh as well as increase the
number of doctors both areas.
It is estimated that each year, 76,000 women die of breast
cancer in South Asia (India, Bangladesh, Nepal, Myanmar,
Pakistan, and Tibet) [2]. In Bangladesh, there is no national
cancer registry but estimate an annual new breast cancer case
burden of 30,000 women. It is projected that global breast
cancer cases will grow from 1.4 million in 2008 to over 2.1
36
International Journal of Computer Applications (0975 – 8887)
Volume 82 – No4, November 2013
million cases in 2030 [3]. Breast Cancer Incidence and
Mortality in Bangladesh will reach an unexpected level. So
it’s time to reduce number of breast cancer patient providing
new technology day after day. At early stage breast cancer
have no symptoms. But after developing or advanced breast
cancer stage some symptoms. Like







Bone pain
Breast pain or discomfort
Skin ulcers
Swelling of one arm
Weight loss
Nipple discharge.
Change in shape or appearance of the nipple
According to breast cancer stage there are different kinds of
treatment. Treatment for breast cancer depends on the type
and stage of the disease, the size the tumor, general health,
medical history and age of patients. In most cases, the goal of
treatment is to remove or destroy the cancer completely. Most
breast cancers can be cured if found and treated early. Like
other diseases treatment, breast cancer treatment has also
some side effects. It may damages healthy cells and tissues,
unwanted side effects sometimes occur. Side effects depend
mainly on the type and extent of the treatment. Side effects
may not be the same for each person. So it tells again that
“Prevention is better than cure”. And this paper will be
helpful to maintain the moral sentence.
In general, cancer treatments may include:
Chemotherapy medicines to kill cancer cells;
Radiation therapy to destroy cancerous tissue; and
But it need more money that not bearable to low income
people of developing country. So it will be a good work to
introduce a system that can detect breast cancer with little
amount and quickly. By thinking those, this paper introduces
a system that can detect breast cancer quickly and need very
low money.
2. BACKGROUND
cancer because of their unconsciousness. This can be removed
or prevented by maintaining some rules.
A person who was affected by any kind of cancer has an
increased risk of developing another cancer of any type [6]. A
person who has two or more close relatives (mother, father,
sister, brother, or child) who are responsible for developing
breast cancer has a risk factor of developing Breast cancer for
his own. Rarely, members of a family will have an inherited
disorder that makes the breast cell also affected and increases
the risk of Breast cancer.
Sometimes some people develop breast cancer by genes
(mostly found BRCA1 and BRCA2 genes). These genes
normally produce proteins that protect us from cancer. If a
parent passes you a defective gene, you have an increased risk
for breast cancer. Women with one of these defects have up to
an 80% chance of getting breast cancer sometime during their
life.
Women who got their periods early (before age 12) or went
through menopause late (after age 55) have an increased risk
for breast cancer. These 2 risk factors are high risk factor of
occurring breast cancer.
Women who have never had children or who had them only
after age 30 have an increased risk for breast cancer. Being
pregnant more than once or becoming pregnant at an early age
reduces your risk of breast cancer.
If any person received radiation therapy as a child or young
adult to treat cancer of the chest area, they have a much higher
risk for developing breast cancer. A person who has taken
any therapy is responsible for occurring different kinds of
disease compare to other people who don’t take any therapy
[2], [6]. Women who took diethylstilbestrol (DES) to prevent
miscarriage may have an increased risk of breast cancer after
age 40.
Abortion is other most important risk factor of breast cancer.
Taking alcohol regularly increase the risk level of breast
cancer. Drinking alcohol increases the risk of breast cancer.
Experts recommend no more than one alcoholic beverage a
day for women.
Breast cancer, second deathful disease in Bangladesh can
detect or predict by analysing electrical conical images, cancer
region images or treatment information. Like Skin Cancer [4],
Lung Cancer [5], Brain cancer [6] and Diabetics [7], this
research paper also introduces to predict Breast cancer using
treatment information. That treatment information has been
collected according to Bangladesh and Bangladeshi patients.
Treatment information is described briefly in below section.
Obesity is not only responsible for diabetics or breast cancer
but also responsible for other diseases forcing directly or
indirectly. Obesity is a condition in which a person has an
abnormally high and unhealthy proportion of body fat. To
measure obesity mainly used body mass index (BMI) that
calculated by dividing a person’s weight (in kilograms) by
their height (in meters) squared. It provides a more accurate
measure of obesity or being overweight than weight alone.
Determining what causes cancer is complex. Many things are
known to increase the risk of cancer, that are called risk
factors of that cancer, including tobacco use, certain
infections, radiation, lack of physical activity, obesity, and
age. If any person is aware about these risk factors then cancer
can prevent easily. Like of other cancers breast cancer also
depends on some factors where some are most important and
some are less important. Also some are editable and some are
not editable. In this section discussed about the risk factors of
breast cancer as well as their level of importance.
The diagnosis of Breast Cancer is a tedious and significant
task. Advances in breast cancer treatment are facilitated by the
knowledge gained from tumor biopsy and pathology. These
techniques allow doctors to develop individual treatments for
each patient based on their distinctive tumor biology. New
techniques are constantly being researched to improve the
detection, diagnosis, and treatment of breast cancer likely
brain cancer, skin cancer [4], Lung cancers [5], Diabetics [7].
Age and gender is the most important factors of causing breast
cancer. Compare to men, women are mainly affected by this
cancer and developing of breast cancer risk is increased
according to becoming older. In Bangladesh more than 50
years old people are mainly affected by breast cancer.
Unmarried younger girl are also affecting rapidly by this
The treatment of breast cancer risk factors can be broken
down into two categories: removable and irremovable risk
factors. Removable risk factors are those risk factors that can
be solved by consciousness or taking some rules. Irremovable
risk factors are those risk factors that can’t solve but keep
those factors unworkable. Table 2 shows the categories of
breast cancer risk factors.
37
International Journal of Computer Applications (0975 – 8887)
Volume 82 – No4, November 2013
A widely recognized formal definition of data mining can be
defined as “Data mining is the non- trivial extraction of
implicit previously unknown and potentially useful
information about data” [8]. Data mining has some fields to
analysis of data such as classification, clustering, correlations,
association rule etc. Now-a-days data mining has been used
intensively and extensively by many organizations. InHealthcare, data mining is becoming increasingly popular [9].
Data mining provides the methodology and technology to
analysis the useful information of data for decision making.
Pre-processed data are broken down into two categories:
relevant and non-relevant dataset to breast Cancer. Clustering
technique mainly used to find out two types of data.
Clustering is a process of separating dataset into subgroups
according to the unique feature. WEKA toolkits are mainly
used to find out highest affected risk factors as well as ranking
of risk factors from dataset.
3.3 Discover Most Responsible Factors
Day by day the number of Breast cancer person is increasing
rapidly because of unconsciousness about breast cancer and
risk factors of breast cancer. The main goal of this research is
to develop a system that can be used by a person for testing
his/her breast Cancer risk level and provides suggestions
according to his/her breast cancer risk factor.
The experimental results are categorized into two sections.
One is ranking factors discover and another is represents
prediction tools to Breast Cancer.
3. METHODOLOGY
3.1 Data Collection and Pre-processing
100 persons’ data is collected from different diagnostic centre
where only female are present of age 18 to 70. Obtained data
also contains both patient and non-patient information. From
the previous studies 20 risk factors were considered for Breast
cancer assessment in Bangladeshi population, which includesage, gender, hereditary, previous health examination, use of
anti-hypersensitive drugs, late Menopause, early Menarche,
food habit, obesity, genetic risk, environment, excessive
alcohol, hormone therapy, abortion ,exercise, Depression,
marital status, outdoor activities and affected any cancer
before.
Discover most responsible of risk factors is known as the
principle data mining problem that intends to find out the
frequent items from the data warehouse. Here CFS Subset
Evaluator Including locally predictive attributes in WEKA
toolkit is used to discover the highest responsible factors of its
efficiency and simplicity than other algorithms where relevant
data in CSV format to Breast Cancer are used as input of this
algorithm.
3.4 Ranking Risk Factors
Here Chi-squared Ranking Filter is used as Attribute
Evaluator and Ranker is used as search method in WEKA to
rank 13 included risk factors because of its efficiency and
simplicity than others.
4. RESULT
4.1 Results for Ranking Risk Factors
The ranking of Breast cancer risk factors and highest
responsible factors of causing breast cancer are shown in Fig.
1 and Fig. 2 respectively. Here we assign lowest factors score
as 1 and highest score as 4. Each factor is subdivided into
different sectors and assigns a different score that is presented
in Table 1 that is used for predicting Breast cancer.
4.2 Results for Prediction to Breast Cancer
Finally using Table 1 the prediction tool to breast Cancer is
implemented that is shown in Fig. 3. Fig. 4 represents the risk
level (flowchart) of breast Cancer which is implemented using
Table 1. In Table 1 weightage score high means high impact
on causing breast cancer. Fig. 3 is principle graphical
representation and Fig. 5-6, 8-9 is the graphical representation
of Breast cancer different risk levels as example.
3.2 Pre-processed of Obtain Data
Due to missing and duplicate information the collected data
are pre-processed by avoiding duplicate and supplying
missing information. It mainly reduces the memory and
normalizes the values used to represent information in
database. And after analysis which factors are not link with
breast cancer assign risk value to zero (0) and others are link
with breast cancer assign to risk value one (1). Now classify
the factors into 2 groups one contains relevant data to Breast
Cancer who has risk value 1 and another contains remaining
data that means who has risk value 0.
The process of partitioning and category of collected data into
different subgroups where each groups have a unique feature
is called clustering [10]. The goal of clustering is to classify
objects or data into a number of categories or classes where
each class contains identical feature. The main benefits of
clustering are that the data object is assigned to an unknown
class that have unique feature and reduces the memory. The
K-means clustering [11] is a widely recognized clustering tool
that is used for robotics, diseases and artificial intelligence
application purposes [12]. Here k is a positive integer
representing the number of clusters. The pre-processed data is
clustered using the K-means clustering algorithm with the
value of k equal to 2. This represents there is two clusters
where one cluster contains relevant data to Breast Cancer and
another contains remaining data.
Fig 1: Most affected risk factors list using WEKA
5. COMPARE TO EXIT TECHNIQUES
Through Mammography [12] breast cancer can detect but it
need more time, cost and increase radiation problem because
38
International Journal of Computer Applications (0975 – 8887)
Volume 82 – No4, November 2013
of X-ray. MRI has a role at high risk for breast cancer of
women. Latest diagnostic techniques such as sestamibi scans,
optical imaging and molecular diagnostic techniques [13] look
promising, but need more investigation and time into their
use. PET/computed tomography [14] a role in detecting local
disease recurrence and distant metastasis in breast cancer
patients but not time savable as well as cost. But this paper
introduces a new technique that is easy, secure, time savable
as well as cost savable.
Table 1. Factors corresponding Weightage & Score
Parameters
Weight age
Score
Age
Age≤24
24<Age≤38
38<Age≤50
Age>50
Yes
No
No
Yes
No
Uncle, Aunty
Brother, Parents
etc.
BMI≤24
25≤BMI≤28
BMI≥29
Yes
No
No
Yes
Yes
No
Yes
No
Yes
No
Yes
No
Yes
No
1
2
3
4
1
2
1
4
1
2
5
Married?
Menarche started early
(before 12 years)?
Affected Family
members?
BMI?
Menopause stops at late
(after 53 years)?
Fatty foods Habits?
Alcohol Use?
Hormone Therapy in
Breast Area?
Affected any cancer
before?
Abortion
Exercise Regularly?
6. CONCLUSION
Early detection of breast cancer is imperative for improving
patient survival. Breast cancer ranks as one of the leading
cancer types in the number of new cases diagnosed and is
second only to lung cancer as the most prevalent cause of
cancer death in women. For treatment it needs more money
that is not bearable for developing countries’ people like
Bangladeshi people. Due to Bangladesh our technique is
constantly being researched to improve the detection,
diagnosis, and treatment of breast cancer.
Table 2. Removable and irremovable risk factors
Removable Risk
Factors
Irremovable Risk
Factors
1. Marital Status;
2. BMI;
3. Early Menarche;
4. Late Menopause;
5.Hormone therapy;
6. Alcohol Use;
7. Fatty foods;
8. Abortion and
9. Regular Exercise
1. Age;
2. Sex;
3. Affected family
members;
4. Affected any cancer
before;
1
2
3
4
1
1
2
3
1
3
1
4
1
3
1
1
2
Fig 4: ‎Flowchart Breast cancer risk prediction system
Fig 2: Ranking of risk factors using WEKA
Fig 3: Graphical view of Breast cancer risk prediction
39
International Journal of Computer Applications (0975 – 8887)
Volume 82 – No4, November 2013
Fig 5‎: Graphical view of Breast Cancer Prediction with
Low Risk Level
Fig 6‎: Graphical view of Breast Cancer Prediction with
Intermediate Risk Level
Fig 7: Visualization of Breast Cancer data where Red & Blue color represent patient & non-patient respectively
40
International Journal of Computer Applications (0975 – 8887)
Volume 82 – No4, November 2013
Fig 8: Graphical view of Breast Cancer Prediction with
High Risk Level
7. ACKNOWLEDGMENTS
The authors are grateful to the participants who contributed to
this research.
8. REFERENCES
[1] H. L. Story, R. R. Love, R. Salim, A. J. Roberto, J. L.
Krieger, O. M., “Improving Outcomes from Breast
Cancer in a Low-Income Country: Lessons from
Bangladesh”, International Journal of Breast Cancer,
2012, pp. 1-9. Article ID 423562.
[2] M. Brown, S. Goldie, G. Draisma, J. Harford, J. Health
service interventions for cancer control in developing
countries in Disease Control Priorities in Developing
Countries. Oxford University Press, 2006, vol. 2, pp.
569-90.
[3] World Cancer Research Fund. Available
http://www.wcrf.org/cancer_facts/women-breastcancer.php/. Accessed March 10, 2013.
from:
[4] K. Ahmed, T. Jesmin, Md. Z. Rahman, “Early
Prevention and Detection of Skin Cancer Risk Using
Data Mining” International Journal of Computer
Applications, 2013, vol. 62, pp. 1-6.
[5] K. Ahmed, Abdullah-Al-Emran, T. Jesmin, R. F. Mukti,
Md. Z. Rahman, F. Akter, “Early Detection of Lung
Cancer Risk Using Data Mining”, Asian Pacific Journal
of Cancer Prevention, 2013, vol. 14 pp. 595-98.
[6] T. Jesmin, K. Ahmed, Md. Z. Rahman, Md. B. A. Miah,
“Brain Cancer Risk Prediction Tool Using Data Mining”,
International Journal of Computer Applications, 2013,
vol.61, pp. 22-27.
IJCATM: www.ijcaonline.org
Fig 9‎: Graphical view of Breast Cancer Prediction with
Very High Risk Level
[7] K. Ahmed, T. Jesmin, U. Fatima, Md. M., Abdullah-alE., Md. Z. Rahman, “Intelligent and Effective Diabetes
Prediction System Using Data Mining Approach”,
ORIENTAL JOURNAL OF COMPUTER SCIENCE &
TECHNOLOGY, 2012, vol. 5, pp. 215-21.
[8] Frawley, Piatetsky S., “Knowledge Discovery in
Databases: An Overview”, the AAAI/MIT Press Menlo
Park C.A.
[9] H. C. Koh, G. Tan, “Data Mining Applications in
Healthcare” Journal of Healthcare Information
Management, vol. 19, pp. 64-72.
[10] Z. Nouir, B. Sayrac, B. Fourestié, W. Tabbara, F. B.,
“Generalization Capabilities Enhancement of a Learning
System by Fuzzy Space Clustering” Journal of
Communications, 2007, vol. 2, pp. 30-7.
[11] C. Ordonez, “Programming the K-Means Clustering
Algorithm in SQL”, in Proc. ACM Int’l Conf.
Knowledge Discovery and Data Mining, 2004, pp. 8238.
[12] Daniel B. K., “Early Breast Cancer Detection Using
Techniques other than Mammography”, American
Journal of Roentgenology, 1984, pp. 465-8.
[13] Medical Center University of Bonn, “Improving
Methods for Breast Cancer Detection and Diagnosis”,
National Cancer Institute, 2003.
[14] S. V, S. C, W. L, Bourke A., “New diagnostic techniques
for breast cancer detection” US National Library of
Medicine National Institutes of Health, 2008, vol. 4, pp.
501-13.
41